Rekh Ram Janghel

Diagnosis of breast cancer by modular evolutionary neural networks

We construct a mixture of experts model for medical diagnosis. Each of the experts is a complex modular neural network. The first modularity clusters the entire input space into a set of modules. The second modularity divides the number of attributes. Each cluster is a neural network that solves the problem. The individual neural networks are evolved using genetic algorithms, which optimise the architecture along with the weights and biases. The complete system is used for the diagnosis of breast cancer. Experimental results show that the proposed system outperforms the traditional simple and hybrid approaches.

Hybrid computing based intelligent system for breast cancer diagnosis

Breast cancer is one of the major causes of death in women which accounts one out of eight. As primary cause is still unknown, early detection increases better treatment and improves total recovery. We present some novel hybrid approaches for classification of breast cancer. Artificial Neural Network (ANN) which suffers credit assignment problem can be avoided by modular and evolutionary artificial neural network which achieves simple and small individual neural network. Ensemble of ANN is to obtain a more reliable and accurate ANN. Evolutionary Neural Network (ENN) is used for optimization of neural network learning and design. The best accuracies achieved for diagnosis are around 99% using breast cancer datasets.

Dimensionality Reduction-Based Breast Cancer Classification Using Machine Learning

In the field of medical science, achieving accurate diagnosis of disease before its treatment is a significant obstacle. A lot of tests are available, which not only complicates the diagnostic process but also finds difficulty in deriving results. Therefore, computational diagnostic techniques must be introduced with the support of artificial intelligence and machine learning. Breast cancer, being one of the second-leading cause of deaths in women worldwide, demands terminal diagnosis with the higher degree of accuracy. In this proposed work, the primary purpose is to establish and contrast the integrated approaches involving dimensionality reduction, feature ranking, fuzzy logic, and neural networks for the diagnostic evaluation of breast cancer, namely, benign and malignant. However, the adopted approach has been successful in giving the optimal performance to a greater extent, but a maximum accuracy of 96.58% is obtained by the use of principal component analysis and backpropagation neural network.

Noise Removal from Epileptic EEG signals using Adaptive Filters

Electroencephalography (EEG) is a well-established clinical procedure which provides information pertinent to the diagnosis of various brain disorders. EEG waves are highly vulnerable to diverse forms of noise which pose notable challenges in the analysis of EEG data. In this paper, adaptive filtering techniques, namely, Recursive Least Squares (RLS), Least Mean Squares (LMS), and Shift Moving Average (SMA) filters, were applied to the collected EEG signals to filter noise from the EEG signal. Various fidelity parameters, namely, Mean Square Error (MSE), Maximum Error (ME), and Signal-to-Noise Ratio (SNR), were observed. Our method has shown better performance compared to previous filtering techniques. Overall, in comparison to the previous methods, this proposed strategy is more appropriate for EEG filtering with greater accuracy.

ECG Arrhythmia Classification Using Artificial Neural Networks

Electrocardiogram (ECG) arrhythmia is referred to as a change in human heart rhythm, and it becomes either too slow or very large compared to normal heart rhythms. This may cause disease affecting cardiac. Early correct identification of arrhythmia is important in the detection of cardiac disease and getting the better treatment of a patient. Numerous classifiers are present for ECG diagnosis. Artificial neural network (ANN) is one of the more popular and very widely utilized models for ECG diagnosis. In this paper, we introduced three different ANN models, which are classified as healthy and arrhythmia classes and using UCI repository ECG 12 lead signal feature extracted data. This particularly uses ANN models that are trained as well as tested on back-propagation feedforward neural network, recurrent neural network (RNN), and radial basis function (RBF) networks. We evaluated the diagnosis testing result in the form of classification accuracy, sensitivity, and specificity. Among these three contrast ANN models, RNN models have shown better diagnosis result up to obtained 83.1% testing classification accuracy with selected attributes.

A comparison of soft computing models for Parkinson’s disease diagnosis using voice and gait features

Parkinson’s disease is a widespread disease among elder population worldwide caused by dopamine loss, which reduces quality of life because of motor and non-motor complications. In the current paper, nine soft computing models, i.e., Cubist, Cubist Committees, Random Forests, Kernel Support Vector Machine, Linear Regression, Naïve Bayes, Artificial Neural Network, Adaptive Neuro-Fuzzy Inference System, and Hybrid Neuro-Fuzzy Inference System are implemented for Parkinson’s disease diagnosis using voice and gait features. Later, their performances are evaluated based on performance measures, viz., true positive, false positive, false negative, true negative, accuracy, sensitivity, specificity, and RMSE, and finally, a comparison is performed to identify the most efficient model and data set combination. The comparison demonstrated that Random Forest model outperformed others yielding 100% accuracy, 100% sensitivity, 100% specificity, and zero RMSE on voice and gait training data sets both; Cubist Committees model outperformed others yielding 74.00% accuracy, 69.39% sensitivity, 78.43% specificity, and 0.4582 RMSE on voice testing data set; Random Forest model once again outperformed others yielding 81.66% accuracy, 92.39% sensitivity, 66.67% specificity, and 0.4283 RMSE on gait testing data set. Furthermore, these models’ performances are also evaluated on reduced feature vector voice and gait data sets obtained by Principal Component Analysis, and compared with their performances on former data sets. The comparison exhibited that the soft computing models’ performance decreases by reducing feature vector of the data sets.

Soft computing based expert system for Hepatitis and liver disorders

The main objective of this research work is to develop an expert system for the diagnosis and detection of Hepatitis and liver disorders based on various Artificial Neural Networks models. In this research work Artificial Neural Networks models like Back Propagation Algorithm, Probabilistic Neural Networks, Competitive learning Networks, Learning vector quantization and Elman Networks have been used for detection and diagnosis of Hepatitis and liver disorders. The various networks developed with the help of MATLAB. Required data has been chosen from trusty machine learning data base (UCI). This system in comparison with other traditional diagnostic systems is faster, more reliable and more accurate. One can use this system as a specialist assistant or for training medicine students.